WO2019190284A1

WO2019190284A1 - Method and apparatus for performing low-complexity operation of transform kernel for video compression

Info

Publication number: WO2019190284A1
Application number: PCT/KR2019/003743
Authority: WO
Inventors: 구문모; 살레히파메흐디; 김승환; 임재현
Original assignee: 엘지전자 주식회사
Priority date: 2018-03-29
Filing date: 2019-03-29
Publication date: 2019-10-03
Also published as: US20210021871A1

Abstract

The present invention provides a method for restoring a video signal on the basis of a low-complexity transform execution, comprising the steps of: acquiring the transform index of a current block from a video signal, wherein the transform index corresponds to any one of a plurality of transform combinations comprising DST4 and/or DCT4; deriving a transform combination corresponding to the transform index, wherein the transform index comprises a horizontal transform and a vertical transform, and the horizontal transform and the vertical transform correspond to either the DST4 or the DCT4; performing an inverse transform on the current block in the vertical direction by using the DST4; performing an inverse transform on the current block in the horizontal direction by using the DCT4; and restoring the video signal by using the inversely transformed current block.

Description

2019/190284 1 »（： 1 ^ 1 {2019/003743

One

【Specification】

[Invention and name]

Method and device for low complexity operation of transform kernel for video compression

The present invention relates to a method and an apparatus for processing a video signal, and more particularly to memory usage and computational complexity of a discrete sine transform-4 (DST4) and a discrete cosine transform-4 (DCT4) in a transform kernel for video compression. To reduce technology.

Background Art

Next generation video content will be characterized by high spatial resolution, high frame rate and high dimensionality of scene representation. Processing such content will result in a tremendous increase in terms of memory storage, memory access rate, and processing power.

Thus, there is a need to design new coding tools to process next generation video content more efficiently. In particular, when applying transforms, it is necessary to design transforms that are much more efficient in terms of coding efficiency and complexity.

[Detailed Description of the Invention]

[Technical problem]

This paper proposes a low complexity algorithm for transform kernel for video compression.

The present invention provides a DST4 (Discrete Sine Transform-) among transform kernels for video compression. 2019/190284 1 »（： 1 ^ 1 {2019/003743

2

4) and to reduce the memory usage and computational complexity for Discrete Cosine Transform-4 (DCT4).

The present invention intends to propose an encoder / decoder structure for reflecting a new transform design.

Technical Solution

The present invention provides a method of reducing complexity and improving coding efficiency through a new transform design.

The present invention, DST4 (Discrete Sine Transform-4) and DCT4 (Discrete Cosine

Transform-4) is provided as a forward DCT2.

The present invention provides a method of performing DST4 and DCT4 as inverse DCT2.

The present invention provides a method of applying DST4 and DCT4 to a transform configuration group to which MTS (Multiple Transform Selection) is applied.

【Effects of the Invention】

By providing a method of performing Transform-4) as forward DCT2 or inverse DCT2, memory usage and computational complexity can be reduced. In addition, the present invention can perform more efficient coding by applying DST4 and DCT4 to a transform configuration group to which MTS (Multiple Transform Selection) is applied.

As such, the new low complexity algorithm can reduce computational complexity and improve coding efficiency. 2019/190284 1 »（： 1 ^ 1 {2019/003743

3

[Brief Description of Drawings]

1 is an embodiment to which the present invention is applied and shows a schematic block diagram of an encoder in which encoding of a video signal is performed.

2 is a schematic block diagram of a decoder in which decoding of a video signal is performed according to an embodiment to which the present invention is applied.

Figure 3 is an embodiment to which the present invention can be applied, Figure 3A is QT (QuadTree, referred to as' QT '), Figure 3B is BT (Binary Tree, referred to as' BT), Figure 3C is TT 3D is a diagram for describing block division structures by an AT (Asymmetric Tree, hereinafter called 'AT').

4 is a schematic block diagram of a transform and quantization unit 120/130 and an inverse quantization and inverse transform unit 140/150 in an encoder according to an embodiment to which the present invention is applied.

5 is a schematic block diagram of an inverse quantization and inverse transform unit 220/230 in a decoder according to an embodiment to which the present invention is applied.

FIG. 6 is a table illustrating a transform configuration group to which an MTS (Multiple Transform Selecti) is applied as an embodiment to which the present invention is applied.

7 is an embodiment to which the present invention is applied and is a flowchart illustrating an encoding process in which MTS (Multiple Transform Selection) is performed.

8 is a flowchart illustrating a decoding process in which multiple transform selection (MTS) is performed, according to an embodiment to which the present invention is applied.

9 is a flowchart illustrating a process of encoding an MTS flag and an MTS index according to an embodiment to which the present invention is applied.

FIG. 10 illustrates an embodiment to which the present invention is applied and includes an MTS flag and an MTS index. 2019/190284 1 »（1 ^ 1 {2019/003743

4 is a flowchart to explain the decoding process of applying a horizontal transform or a vertical transform to a row or a column based on the above.

11 is a schematic block diagram of an inverse transform unit in a decoder according to an embodiment to which the present invention is applied.

12 is a block diagram for performing inverse transformation based on transformation-related parameters according to an embodiment to which the present invention is applied.

13 is an embodiment to which the present invention is applied and shows a flowchart of performing inverse transformation based on transformation related parameters.

14 is an embodiment to which the present invention is applied and shows an encoding flowchart for performing Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) to a forward DCT2 or an inverse DCT2. .

FIG. 15 illustrates a decoding flowchart of performing Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) with a forward DCT2 or an inverse DCT2 as an embodiment to which the present invention is applied. .

FIG. 6 is an embodiment to which the present invention is applied and shows diagonal elements for a pair of transform block size (N) and right shift amount () when DST4 and DCT4 are performed as forward DCT2.

FIG. 17 illustrates sets of DCT2 kernel coefficients that may be applied to DST4 or DCT4 as an embodiment to which the present invention is applied.

18 illustrates a forward DCT2 matrix generated from a set of DCT2 kernel coefficients that may be applied to DST4 or DCT4 as an embodiment to which the present invention is applied.

19 is a code of an output step for DST4 as an embodiment to which the present invention is applied. 2019/190284 1 »（： 1 ^ 1 {2019/003743

5 Indicates run.

20 shows code execution of an output stage for DCT4 as an embodiment to which the present invention is applied.

FIG. 21 shows the configuration of parameter sets and multiplication coefficients for 0814 and 1X74 when 0814 and 1: 14 are performed in a forward 0012 as an embodiment to which the present invention is applied.

22 shows code execution of a preprocessing step for 0014 as an embodiment to which the present invention is applied.

FIG. 23 shows code execution of a preprocessing step for 0814 as an embodiment to which the present invention is applied.

24 illustrates diagonal elements for a pair of transform block sizers) and right shift amount 4) when 0814 and 1X714 are performed in reverse 0012 as an embodiment to which the present invention is applied.

FIG. 25 shows the configuration of parameter sets and multiplication coefficients for 14 and 1X74 when executed with 0814 and 00X41-reverse 0012 as an embodiment to which the present invention is applied.

FIG. 26 illustrates eight mappings to intra prediction residuals according to an embodiment to which the present invention is applied.

FIG. 27 is a diagram illustrating 1 18 mapping of inter prediction residuals according to an embodiment to which the present invention is applied.

28 is a diagram illustrating a structure of a content streaming system according to an embodiment to which the present invention is applied. 2019/190284 1 »(： 1 ^ 1 {2019/003743

6

[Best form for implementation of the invention]

According to an aspect of the present invention, there is provided a method of reconstructing a video signal based on low complexity conversion, comprising: obtaining a transform index of a current block from the video signal, wherein the transform index is composed of a combination of 0814 and / or DC ^ 4. Corresponds to any one of a plurality of transform combinations; Deriving a transform combination corresponding to the transform index, wherein the transform combination comprises a horizontal transform and a vertical transform, wherein the horizontal transform and the vertical transform correspond to either 0814 or 0014; Performing inverse transform in a vertical direction with respect to the current block using the 0314; Performing inverse transform in a horizontal direction with respect to the current block using the 1 × ^ 4; And reconstructing the video signal using the inversely transformed current block.

In the present invention, the 0814 and / or the 1X714 is characterized in that the execution using the forward 0072 or the reverse direction.

In the present invention, 0814 and / or å) ⁽ 4) is characterized by applying a post-processing matrix ^ and a pre-processing matrix to the forward DCT2 or the reverse 1X72. Here, 2003

Indicate size. 2019/190284 1 »（： 1 ^ 1 {2019/003743

7 In the present invention, when the vertical transformation is 14, the inverse transformation of 14 is applied for each column, and when the horizontal transformation is å ⁾ 0:14, the inverse transformation of 1 ^{) (} 714 ⁾ is applied for each row. .

In the present invention, the transformation combination (horizontal transformation, vertical transformation), 쨘 計 4, 0814), ₍ 0 ^ 4, 0814), (0814, 0014) and (1) <the 4, 1) (4) It is characterized by including. In the present invention, when the current block is an intra predicted residual, the transform combination corresponds to the transform indices 0, 1, 2, and 3.

In the present invention, when the current block is an inter predicted residual, the transform combination corresponds to the transform index 3, 2, 1, 0.

The present invention provides an apparatus for reconstructing a video signal based on low complexity conversion execution, the apparatus comprising: a parser for obtaining a transform index of a current block from the video signal, wherein the transform index is a combination of 0814 and / or DC ^ 4 Corresponds to any one of the configured plurality of transform combinations; A transform unit for deriving a transform combination corresponding to the transform index, performing a reverse transformation in the vertical direction with respect to the current block using 4, and performing a reverse transformation in the horizontal direction with respect to the current block using 1 × 74, Wherein the transform combination comprises a horizontal transform and a vertical transform, wherein the horizontal transform and the vertical transform correspond to either 0814 or DCT4; And a reconstruction unit for reconstructing the video signal using the inversely transformed current block.

[Form for implementation of invention]

Hereinafter, with reference to the accompanying drawings will be described the configuration and operation of the embodiment of the present invention, the configuration and operation of the present invention described by the drawings as an embodiment 2019/190284 1 »（： 1 ^ 1 {2019/003743

8 is to be described, and the technical spirit of the present invention and its core configuration and operation are not limited thereto.

In addition, the terminology used in the present invention is selected as a general term widely used as possible now, in a specific case will be described using terms arbitrarily selected by the applicant. In such a case, since the meaning is clearly described in the detailed description of the relevant part, it should not be simply interpreted by the name of the term used in the description of the present invention, and it should be understood that the meaning of the term should be interpreted. .

In addition, terms used in the present invention may be replaced for more appropriate interpretation when there are general terms selected to describe the invention or other terms having similar meanings. For example, signals, data, samples, pictures, frames, blocks, etc. may be appropriately replaced and interpreted in each coding process. In addition, partitioning, decomposition, splitting, and division may be appropriately replaced and interpreted in each coding process. _

In this document, MTS (Multiple Transform Selection) may refer to a method of performing transformation using at least two or more transform types. This may be expressed as an Adaptive Multiple Transform (AMT) or Explicit Multiple Transform (EMT), and likewise, mtsjdx may be expressed as AMTJdx, EMT_idx, tu_mts_idx, AMT_TU_idx, EMT_TU_idx, transform index, or transform combination index, and the present invention. It is not limited to this expression. 2019/190284 1 »（： 1 ^ 1 {2019/003743

FIG. 1 is a schematic block diagram of an encoder in which encoding of a video signal is performed according to an embodiment to which the present invention is applied.

Referring to FIG. 1, the encoder 100 includes an image splitter 1 10, a transform unit 120, a quantization unit 130, an inverse quantization unit 140, an inverse transform unit 150, a filtering unit 160, It may include a decoded picture buffer (DPB) 170, an inter predictor 180, an intra predictor 185, and an entropy encoder 190.

The image divider 1 10 may divide an input image (or a picture or a frame) input to the encoder 100 into one or more processing units. For example, the processing unit may be a Coding Tree Unit (CTU), a Coding Unit (CU), a Prediction Unit (PU), or a Transform Unit (TU).

However, the terms are only used for the convenience of description of the present invention, the present invention is not limited to the definition of the terms. In addition, in the present specification, for convenience of description, the term coding unit is used as a unit used in a process of encoding or decoding a video signal, but the present invention is not limited thereto and may be appropriately interpreted according to the present disclosure.

The encoder 100 may generate a residual signal by subtracting a prediction signal output from the inter predictor 180 or the intra predictor 185 from the input image signal, and generate the residual signal. Is transmitted to the converter 120.

The transform unit 120 may generate a transform coefficient by applying a transform technique to the residual signal. The conversion process consists of a quadtree square block, a binarytree structure, a ternary tree, or a non-symbol. 2019/190284 1 »（： 1 ^ 1 {2019/003743

It can be applied to blocks (square or rectangle) divided by 10 tree (asymmetric) structure.

The transform unit 120 may perform a transform based on a plurality of transforms (or transform combinations), and this transform scheme may be referred to as MTS (Multiple Transform Selection). The MTS may be called an AMT (Adaptive Multiple Transform) or an EMT (Enhanced Multiple Transform).

The MTS (or AMT, EMT) may refer to a transform scheme performed based on a transform (or transform combinations) adaptively selected from a plurality of transforms (or transform combinations).

The plurality of transforms (or transform combinations) may include the transforms (or transform combinations) described with reference to FIGS. 6 and 26 to 27. In the present specification, the transform or transform type may be expressed as, for example, DCT-Type 2, DCT-II, DCT2, or DCT2.

The converter 120 may perform the following embodiments.

The present invention provides a method of performing Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) as forward DCT2.

The present invention provides a method for performing DST4 and DCT4 with inverse DCT2.

Specific embodiments thereof will be described in more detail herein. 2019/190284 1 »（： 1 ^ 1 {2019/003743

1 1 The quantization unit 130 may quantize the transform coefficients and transmit them to the entropy encoding unit 190, and the entropy encoding unit 190 may entropy-code the quantized signal and output the bitstream.

Although the converter 120 and the quantization unit 130 are described as separate functional units, the present invention is not limited thereto and may be combined into one functional unit. In the case of the inverse quantization unit 140 and the inverse transform unit 150, the same may be combined into one functional unit.

The quantized signal output from the quantization unit 130 may be used, for example, to generate a signal. For example, the quantized signal may recover the residual signal by applying inverse quantization and inverse transformation through inverse quantization unit 140 and inverse transform unit 150 in a loop. A reconstructed signal may be generated by adding the reconstructed residual signal to a prediction signal output from the inter predictor 180 or the intra predictor 185.

Meanwhile, deterioration of the block boundary may occur due to the quantization error generated in the compression process as described above. This phenomenon is called blocking artifacts, which is one of the important factors in evaluating image quality. In order to reduce such deterioration, a filtering process may be performed. Through this filtering process, the image quality can be improved by removing the blocking degradation and reducing the error of the current picture.

The filtering unit 160 applies filtering to the reconstruction signal and outputs it to the reproduction apparatus or transmits the decoded picture buffer to the decoded picture buffer 170. The filtered signal transmitted to the decoded picture buffer 170 may be used as the reference picture in the inter predictor 180. like this, 2019/190284 1 »（： 1 ^ 1 {2019/003743

By using the filtered picture as a reference picture in the inter prediction mode, not only image quality but also encoding efficiency may be improved.

The decoded picture buffer 170 may store the filtered picture for use as a reference picture in the inter prediction unit 180.

The inter prediction unit 180 performs temporal prediction and / or spatial prediction to remove temporal redundancy and / or spatial redundancy with reference to a reconstructed picture. Here, since the reference picture used to perform the prediction is a transformed signal that has been quantized and dequantized in units of blocks at the time of encoding / decoding in the previous time, blocking artifacts or ringing artifacts may exist. have.

Accordingly, the inter prediction unit 180 may interpolate the signals between pixels in sub-pixel units by applying a lowpass filter to solve the performance degradation due to discontinuity or quantization of such signals. Herein, the subpixel refers to a virtual pixel generated by applying an interpolation filter, and the integer pixel refers to an actual pixel existing in the reconstructed picture. As the interpolation method, linear interpolation, bi linear interpolation, and Wiener filter may be applied.

The interpolation filter may be applied to the reconstructed picture to improve the accuracy of the example. For example, the inter prediction unit 180 generates an interpolation pixel by applying an interpolation filter to an integer pixel, and uses an interpolated block composed of interpolated pixels as a prediction block. Yes you can.

On the other hand, the intra prediction unit 185 is the block of the current encoding 2019/190284 1 »（： 1 ^ 1 {2019/003743

13 can refer to a sample around it to instantiate the current block. The intra prediction unit 185 may perform the following process to perform intra prediction. First, a reference sample necessary for generating a prediction signal may be prepared. Then, a prediction signal may be generated using the prepared reference sample. Then, the prediction mode is encoded. In this case, the reference sample may be prepared through reference sample padding and / or reference sample filtering. Since the reference sample has been predicted and reconstructed, there may be a quantization error. Accordingly, the reference sample filtering process may be performed for each prediction mode used for intra prediction to reduce such an error.

A prediction signal generated through the inter predictor 180 or the intra predictor 185 may be used to generate a reconstruction signal or to generate a residual signal. 2 is a schematic block diagram of a decoder in which decoding of a video signal is performed as an embodiment to which the present invention is applied.

Referring to FIG. 2, the decoder 200 includes a parser (not shown), an entropy decoder 210, an inverse quantizer 220, an inverse transformer 230, a filter 240, and a decoded picture buffer (DPB). It may include a decoded picture buffer unit) 250, an inter predictor 260, and an intra predictor 265.

The reconstructed video signal output through the decoder 200 may be reproduced through the reproducing apparatus.

The decoder 200 may receive a signal output from the encoder 100 of FIG. 1, and the received signal may be entropy decoded through the entropy decoding unit 210. 2019/190284 1 »（： 1 ^ 1 {2019/003743

The inverse quantization unit 220 obtains a transform coefficient from the entropy decoded signal using the quantization step size information.

The inverse transform unit 230 inversely transforms the transform coefficient to obtain a residual signal.

Herein, the present invention provides a method of configuring a transform combination for each transform configuration group divided by at least one of a prediction mode, a block size, or a block shape. 230 may perform an inverse transform based on the transform combination constructed by the present invention. In addition, the embodiments described herein may be applied.

The inverse transform unit 230 may perform the following embodiments.

According to an aspect of the present invention, there is provided a method of reconstructing a video signal based on low complexity conversion execution, the method comprising: obtaining a transform index of a current block from the video signal, wherein the transform index comprises a plurality of combinations of DST4 and / or DCT4; Corresponds to any one of the transform combinations; Deriving a transform combination corresponding to the transform index, wherein the transform combination is comprised of a horizontal transform and a vertical transform, wherein the horizontal transform and the vertical transform are the DST4 or the 2019/190284 1 »(： 1/10 公 019/003743

1 5

Corresponds to any one of 0014; Performing inverse transformation in a vertical direction with respect to the current block using the 14; Performing an inverse transform in a horizontal direction with respect to the current block using the £ > And reconstructing the video signal using the inversely transformed current block.

In the present invention, the 14 and / or 1) (the 14 is executed using the forward £> (the 2 or the reverse 1) 0:72).

In the present invention, the 0814 and / or 1X714 is applied to the post-processing matrix and pre-processing matrix in the forward 0012 or the reverse 1) 12

Indicate size.

In the present invention, the inverse transform of 14 is applied for each column when the vertical transform is 14, and the inverse transform of 1X ^ 4 is applied for each row when the horizontal transform is 1X ^ 4.

In the present invention, the transform combination (horizontal transform, vertical transform) includes 必 4, 0 ^ 4), (0014, 0814), (0814, 0014), and (0 4, 1) 14 do. In the present invention, when the current block is an intra predicted residual, the transform combination corresponds to the transform indices 0, 1, 2, and 3. 2019/190284 1 »（： 1 ^ 1 {2019/003743

Although the inverse quantization unit 220 and the inverse transform unit 230 are described as separate functional units, the present invention is not limited thereto and may be combined into one functional unit.

A reconstructed signal is generated by adding the obtained residual signal to a prediction signal output from the inter predictor 260 or the intra predictor 265.

The filtering unit 240 applies filtering to the reconstructed signal and outputs the filtering to the reproducing apparatus or transmits it to the decoded picture buffer unit 250. The filtered signal transmitted to the decoded picture buffer unit 250 may be used as the reference picture in the inter predictor 260.

Seen _. In the specification, the embodiments described in the transform unit 120 and the respective functional units of the encoder 100 may be equally applied to the inverse transform unit 230 and the corresponding functional units of the decoder, respectively. 3 is an embodiment to which the present invention may be applied, FIG. 3A is a QT (QuadTree, hereinafter referred to as QT), FIG. 3B is a Binary Tree (BT), and FIG. 3C is a TT (Ternary Tree). 3D is a diagram for describing block division structures by an AT (Asymmetric Tree, hereinafter called 'AT').

In video coding, one block may be divided on a QT (QuadTree) basis. In addition, one sub-block divided by QT uses QT. 2019/190284 1 »（： 1 ^ 1 {2019/003743

17 can be further divided recursively. A leaf block that is no longer QT split may be split by at least one of a binary tree (BT), a primary tree (TT), and an asymmetric tree (AT). BT may have two types of divisions: horizontal BT (2NxN, 2NxN) and vertical BT (Nx2N, Nx2N). The TT may have two types of divisions, horizontal TT (2Nxl / 2N, 2NxN, 2Nxl / 2N) and vertical TT (l / 2Nx2N, Nx2N, l / 2Nx2N). AT is horizontal-up AT (2Nxl / 2N, 2Nx3 / 2N), horizontal-down AT (2Nx3 / 2N, 2Nxl / 2N), vertical-left AT (l / 2Nx2N, 3 / 2Nx2N), vertical-right AT (3 / 2Nx2N, l / 2Nx2N) can be divided into four types. Each BT, TT, or AT may be further recursively divided using BT, TT, and AT.

3A shows an example of QT division. Block A may be divided into four sub-blocks (AO, Al, A2, A3) by QT. The sub-block time may be divided into four sub-blocks B0, B1, B2, and B3 by Ah.

3B shows an example of BT partitioning. (B2, which is no longer divided by the group, can be divided into vertical BT (CO, Cl) or horizontal BT (DO, Dl). Each sub-block, like block C0, is horizontal BT (E0, El) or vertical. It can be further recursively divided into the form of BT (F0, F1).

3C shows an example of TT partitioning. Block B3, which is no longer split by QT, may be split into vertical TT (CO, Cl, C2) or horizontal TT (DO, Dl, D2). Like the block Cl, each subblock may be further recursively divided into the form of horizontal TT (E0, El, E2) or vertical TT (F0, F1, F2).

Mall 3D shows an example of AT splitting. Block B3, which is no longer split by Ah, can be split into vertical AT (CO, Cl) or horizontal AT (DO, Dl). 2019/190284 1 »（： 1 ^ 1 {2019/003743

Each block can be further recursively divided into the form of horizontal AT (E0, El) or vertical TT (F0, F1). On the other hand, BT, TT, AT partitions can be used together to divide. For example, a subblock divided by BT may be divided by TT or AT. In addition, the sub-block divided by TT can be divided by BT or AT. A subblock divided by AT may be divided by BT or TT. For example, after horizontal BT partitioning, each subblock may be divided into vertical BTs, or after vertical BT partitioning, each subblock may be split into horizontal BTs. Are different, but the final split is the same.

In addition, when a block is divided, the order of searching for the block may be variously defined. In general, searching from left to right, from top to bottom, and searching for a block means an order of determining whether or not to further divide each divided sub-block, or when each sub-block is no longer divided A coding order of a block may be referred to, or a search order when referring to information of another neighboring block in a subblock. 4 and 5 are embodiments to which the present invention is applied, and FIG. 4 is a schematic block diagram of a transform and quantization unit 120/130 and an inverse quantization and inverse transform unit 140/150 in an encoder, and FIG. 5. Shows a schematic block diagram of the inverse quantization and inverse transform units 220/230 in the decoder.

Referring to FIG. 4, the transform and quantization unit 120/130 is a primary transform unit (primary). 2019/190284 1 »（： 1/10 公 019/003743

19 transform unit 121, a secondary transform unit 122, and a quantization unit 130. The inverse quantization and inverse transform unit 140/150 may include an inverse quantization unit 140, an inverse secondary transform unit 151, and an inverse primary transform unit 152. Can be.

Referring to FIG. 5, the inverse quantization unit 220/230 includes an inverse quantization unit 220, an inverse secondary transform unit 231, and an inverse primary transform unit. 232 may include.

In the present invention, the transformation may be performed through a plurality of steps when performing the transformation. For example, two stages of a primary transform and a secondary transform may be applied as shown in FIG. 4, or more transformation steps may be used according to an algorithm. Here, the primary transform may be referred to as a core transform.

The primary transform unit 121 may apply a primary transform to the residual signal, wherein the primary transform is set in the table at the encoder and / or the decoder. Can be

In the case of the primary transform, Discrete Cosine Transform type 2 (hereinafter referred to as "DCT2") may be applied, or Discrete Sine Transform-type 7 (hereinafter referred to as "DST7") in a specific case. For example, DST7 may be applied to a 4x4 block in the intra prediction mode.

In addition, in the case of the primary transform, combinations of various transforms (DST 7, DCT 8, DST 1, DCT 5) of MTS (Multiple Transform Selection) may be applied. For example, Figure 6 can be applied. 2019/190284 1 ^» （： 1 ^ 1 {2019/003743

20 The secondary transform unit 122 may apply a secondary transform on the primary transformed signal, where the secondary transform is an encoder and / or a decoder. Can be defined as in the table.

In an embodiment, the secondary transform may be conditionally applied to a non-separable secondary transform (hereinafter, referred to as 'NSST'). For example, the NSST is applied only to an intra prediction block and may have a transform set applicable to each prediction mode group.

Here, the prediction mode group may be set based on symmetry with respect to the prediction direction. For example, since the prediction mode 52 and the prediction mode 16 are symmetric with respect to the prediction mode 34 (diagonal direction), the same transform set may be applied by forming one group. In this case, when the transform for the prediction mode 52 is applied, the input data is transposed and then applied, since the prediction set 16 and the transform set are the same.

Meanwhile, in the planar mode and the DC mode, since there is no symmetry of the direction, each plane has its own transform set, and the transform set may be composed of two transforms. For the remaining directional mode, three transforms may be configured per transform set.

In another embodiment, the NSST may not be applied to the entire first transformed block but may be applied only to a top-left 8 × 8 region. For example, if the block size is 8x8 or more, 8x8 NSST is applied, and if the block size is less than 8x8, 4x4 NSST7> is applied.

In another embodiment, even when 4xN / Nx4 (N (= 16), 4x4 NSST may be applied. () 2019/190284 1 »(： 1/10 公 019/003743

21 The quantization unit 130 may perform quantization on the quadratic transformed signal. The inverse quantization and inverse transform unit 140/150 performs the above-described process in reverse, and redundant description thereof will be omitted. 5 shows a schematic block diagram of inverse quantization and inverse transform units 220/230 in a decoder.

The inverse secondary transform unit 231 performs inverse secondary transform on the transform coefficients. Here, the inverse secondary transform indicates an inverse transform of the secondary transform described with reference to FIG. 4.

The inverse primary transform unit 232 performs an inverse first transform on an inverse secondary transformed signal (or block) and obtains a residual signal. Here, the inverse first transform represents an inverse transform of the primary transform described with reference to FIG. 4.

The present invention provides a method of configuring a transform combination for each transform configuration group divided by at least one of a prediction mode, a block size, or a block shape. An inverse primary transform unit 232 may be used for transform combinations constructed by the present invention. 2019/190284 1 »（： 1 ^ 1 {2019/003743

Inverse transformation can be performed on the basis of the following. In addition, the embodiments described herein may be applied. FIG. 6 is a table illustrating a transform configuration group to which Multiple Transform Selection (MTS) is applied as an embodiment to which the present invention is applied.

Transformation group to which MTSf Multiple Transform Selection is applied

In the present specification, the j th transform combination candidate for the transform setting group Gi is represented by a pair as shown in Equation 1 below.

[Equation 1]

(H (GiJ), V (GiJ))

Here, H (Gi, j) indicates a horizontal transform for the j th candidate, and V (Gi, j) indicates a vertical transform for the j th candidate. For example, in FIG. 6, H (G ₃ , 2) = DST7, V (G _3J 2)-DCT8 can be expressed. Depending on the context, the value assigned to H (Gi, j) or V (Gi, j) may be a nominal value to distinguish between transformations, as in the example above, or may be an index value indicating the transformation, or It may be a 2D matrix for the transformation.

In addition, in the present specification, 2D matrix values for DCT and DST may be expressed as in Equations 2 to 3 below.

[Equation 2]

DCT type 2:, DCT type 8: Cf ^v

[Equation 3] 2019/190284 1 »（： 1 ^ 1 {2019/003743

23

DST type 7: S _N ^V ", DST type 4: S _N ^{! V}

Here, whether DST or DCT is represented by S or C, the type number is represented by a superscript in the form of Roman numerals, and the N in the subscript indicates NxN conversion. In addition, 2D matrices such as Cl and S _N ^JV assume that column vectors form a transform basis. Referring to FIG. 6, transform configuration groups may be determined based on a prediction mode, and the number of groups may be six (G0 to G5). In addition, G0 to G4 correspond to a case where intra prediction is applied, and G5 represents transform combinations (or transform sets or transform combination sets) applied to a residual block generated by inter prediction.

One transform combination is a horizontal transform (or row transform) applied to the rows of the corresponding 2D block and a vertical transform (or column) applied to the columns. It can consist of a column (column transform).

Here, all of the transform configuration groups may have four transform combination candidates. The four transform combination candidates may be selected or determined through a transform combination index of 0 to 3, and may encode and transmit the transform combination index from an encoder to a decoder.

In one embodiment, the residual data (or residual signal) obtained through intra prediction may have different statistical characteristics according to the intra prediction mode. Thus, the above figure

For intra prediction modes, as in 6, a different cosine transform 2019/190284 1 »（： 1 ^ 1 {2019/003743

24 transformations can be applied.

Referring to FIG. 6, a case of using 35 intra prediction modes and a case of using 67 intra prediction modes are illustrated. A plurality of transform combinations may be applied to each transform setting group divided in each intra prediction mode column. For example, the plurality of transformation combinations may be composed of four (row direction transformation, column direction transformation) combinations. As a specific example, in Group 0, DST-7 and DCT-5 can be applied in both the row (horizontal) direction and the column (vertical) direction, so a total of four combinations are possible.

Since a total of four transform kernel combinations may be applied to each intra prediction mode, a transform combination index for selecting one of them may be transmitted for each transform unit. In the present specification, the transform combination index may be referred to as an MTS index and may be expressed as mts_idx.

In addition to the conversion kernels shown in FIG. 6, a case where DCT2 is optimal for both the row direction and the column direction may occur due to the characteristics of the residual signal. Therefore, the transformation can be adaptively performed by defining the MTS flag for each coding unit. Here, if the MTS flag is 0, DCT2 is applied to both the row direction and the column direction. If the MTS flag is 1, one of four combinations may be selected or determined through the MTS index.

In an embodiment, when the MTS flag is 1, if the number of non-zero transform coefficients for one transform unit is not greater than a threshold value, the DST− for both the row direction and the column direction is not applied without applying the transform kernels of FIG. 6. 7 can be applied. For example, the threshold may be set to 2, which may be set differently based on the block size or the size of the transform unit. This also applies to other embodiments of the specification. 2019/190284 1 »（： 1 ^ 1 {2019/003743

25 is possible.

In one embodiment, if the number of non-zero transform coefficients is parsed by first parsing transform coefficient values, the amount of additional information transmission may be reduced by applying DST-7 without parsing an MTS index.

In an embodiment, when the MTS flag is 1, when the number of non-zero transform coefficients for one transform unit is greater than the threshold, the MTS index may be parsed, and the horizontal transform and the vertical transform may be determined based on the MTS index.

In one embodiment, the MTS may be applied only when the width and height of the transform unit are both 32 or less.

In an embodiment, FIG. 6 may be preset through off-line training.

In one embodiment, the MTS index may be defined as one index that can simultaneously indicate a combination of a horizontal transform and a vertical transform. Alternatively, the MTS index may separately define a horizontal transform index and a vertical transform index.

In one embodiment, the MTS flag or the MTS index may be defined at at least one level of a sequence, picture, slice, block, coding unit, transform unit, or prediction unit. For example, the MTS flag or the MTS index may be defined at at least one level of a sequence parameter set (SPS) or a conversion unit. 7 is an embodiment to which the present invention is applied and is a flowchart illustrating an encoding process in which MTS (Multiple Transform Selection) is performed. 2019/190284 1 »（： 1/10 公 019/003743

In this specification, an embodiment in which transforms are applied to the horizontal direction and the vertical direction is basically described, but the transform combination may be configured as non-separable transforms.

Alternatively, it may consist of a mixture of separable and non-separable transforms. In this case, if a non-separable transform ⁰ ] is used, the selection of transforms by row / column or selection by horizontal / vertical direction is unnecessary and a separable transform is required. The transformation combinations of FIG. 6 above can be used only when.

In addition, the schemes proposed in this specification may be applied regardless of a primary transform or a secondary transform. That is, there is no restriction that it should be applied to either one, and both can be applied. Here, the primary transform may mean a transform for transforming a residual block first, and the secondary transform is applied to a block generated as a result of the primary transform. This may mean a transformation for applying a transformation to the.

First, the encoder may determine a transform setting group corresponding to the current block (S710). Here, the conversion setting group may mean the conversion setting group of FIG. 6, but the present invention is not limited thereto and may be configured with other conversion combinations.

The encoder may perform transform on candidate transform combinations available in the transform configuration group (S720).

As a result of the conversion, the encoder has the lowest RD (Rate Distortion) cost. 2019/190284 1 »（： 1 ^ 1 {2019/003743

27 A combination of transformations may be determined or selected (S730).

The encoder may encode a transform combination index corresponding to the selected transform combination (S740). 8 is a flowchart illustrating a decoding process in which MTS (Multiple Transform Selection) is performed according to an embodiment to which the present invention is applied.

First, the decoder may determine a transform setting group for the current block (S810). The decoder may parse (or obtain) a transform combination index from a video signal, where the transform combination index may correspond to any one of a plurality of transform combinations in the transform configuration group (S820). For example, the transform configuration group may include Discrete Sine Transform type 7 (DST7) and Discrete Cosine Transform type 8 (DCT8). The transform combination index may be referred to as an MTS index. In one embodiment, the transform setting group may be set based on at least one of a prediction mode, a block size, or a block shape of the current block.

The decoder may derive a transform combination corresponding to the transform combination index (S830). The transform combination may include a horizontal transform and a vertical transform, and may include at least one of the DST-7 and the DCT-8.

Also, the transform combination may mean the transform combination described with reference to FIG. 6, but the present invention is not limited thereto. That is, the configuration by other conversion combinations according to another embodiment of the present specification is also possible.

The decoder may perform an inverse transform on the current block based on the transform combination (SM0). The transformation combination converts rows (horizontal) and columns (vertical) 2019/190284 1 »（： 1 ^ 1 {2019/003743

If it consists of 28 transformations, you can apply the row (horizontal) transformation first and then apply the column (vertical) transformation. However, the present invention is not limited thereto, and in the case of applying the reverse or non-separated transform, the non-separated transform may be applied immediately.

In an embodiment, when the vertical transform or the horizontal transform is 081-7 or, the inverse transform of 0 -7 or 0 (the inverse transform of -8 may be applied for each row after applying the column for each column.

In one embodiment, the vertical transformation or the horizontal transformation, a different transformation may be applied to each row and / or each column.

In one embodiment, the transform combination index may be obtained based on a yaw flag indicating whether 1 18 is performed. That is, the transform combination index is assigned to the MTS flag.

May be obtained if performed.

In one embodiment, the decoder may determine whether the number of non-zero transform coefficients is greater than a threshold. In this case, the transform combination index may be obtained when the number of non-zero transform coefficients is greater than a threshold.

In one embodiment, the MTS flag or the

An index may be defined at at least one level of a sequence, picture, slice, block, coding unit, transform unit, or prediction unit.

In one embodiment, the inverse transform may be applied only when the width and height of the transform unit are both 32 or less.

On the other hand, in another embodiment, the process of determining the transform set group and the process of parsing the transform combination index may be performed simultaneously. Alternatively, step 8810 may be omitted and preset in the encoder and / or decoder. 2019/190284 1 »（： 1 ^ 1 {2019/003743

29

The encoder may determine whether MTS (Multiple Transform Selection) is applied to the current block (S910).

If MTS (Multiple Transform Selection) is applied, the encoder can encode MTS flag = 1 (S920).

The encoder may determine an MTS index based on at least one of a prediction mode, a horizontal transform, and a vertical transform of the current block (S930). Here, the MTS index means an index indicating any one of a plurality of transform combinations for each intra prediction mode, and the MTS index may be transmitted for each transform unit.

If the MTS index is determined, the encoder can encode the MTS index (S940).

On the other hand, when the MTS (Multiple Transform Selection) is not applied, the encoder can encode the MTS flag = 0 (S950). FIG. 10 is a flowchart illustrating a decoding process of applying a horizontal transform or a vertical transform to a row or a column based on an MTS flag and an MTS index as an embodiment to which the present invention is applied.

The decoder may parse the MTS flag from the bitstream (S1010). Here, the MTS flag is set to MTS (Multiple Transform Selection) for the current block. \\) 2019/190284 1 »（： 1 ^ 1 {2019/003743

30 may indicate whether or not applied.

The decoder may determine whether MTS (Multiple Transform Selection) is applied to the current block based on the MTS flag (S1020). For example, it may be checked whether the MTS flag is 1 or not.

If the MTS flag is 1, the decoder may check whether the number of non-zero transform coefficients is greater than (or greater than) a threshold value (S1030). For example, the threshold may be set to 2, which may be set differently based on the block size or the size of the transform unit.

If the number of non-zero transform coefficients is larger than a threshold value, the decoder may parse an MTS index (S1040). Here, the MTS index means an index indicating any one of a plurality of transform combinations for each intra prediction mode or inter prediction mode, and the MTS index may be transmitted for each transform unit. Alternatively, the MTS index may mean an index indicating any one transform combination defined in a preset transform combination table, and the preset transform combination table may mean the FIG. 6, but the present invention is limited thereto. It doesn't work.

The decoder may derive or determine a horizontal transform and a vertical transform based on at least one of the MTS index and the prediction mode (S1050).

Alternatively, the decoder may derive a transform combination corresponding to the MTS index. For example, the decoder may derive or determine a horizontal transform and a vertical transform corresponding to the MTS index.

On the other hand, when the number of non-zero transform coefficients is not greater than a threshold value 2019/190284 1 »（： 1 ^ 1 {2019/003743

31 The decoder may apply a predetermined vertical inverse for each column (060). For example, the vertical inverse transform may be an inverse transform of 17.

The decoder may apply a predetermined horizontal inverse transformation for each row 1070. For example, the horizontal inverse transform may be an inverse transform of 17. That is, when the number of non-zero transform coefficients is not greater than a threshold value, a transform kernel preset by an encoder or a decoder may be used. For example, a conversion kernel that is widely used may not be defined in the conversion combination table as shown in FIG. 6.

On the other hand, if the 1 18 flag is 0, the decoder may apply a predetermined tube vertical inverse transform for each column 1080). For example, the vertical inverse transform may be an inverse transform of 1 ⁾ 12.

In addition, the decoder may apply a predetermined horizontal inverse transformation for each row 1090. For example, the horizontal inverse transform may be an inverse transform of 0012.

When the flag is 0, a conversion kernel preset at the encoder or the decoder may be used. For example, a conversion kernel that is widely used may not be defined in the conversion combination table as shown in FIG. 6. FIG. 1 is a schematic block diagram of an inverse transform unit in a decoder according to an embodiment to which the present invention is applied.

The decoder to which the present invention is applied includes a second inverse transform application determining unit (or an element for determining whether to apply the second inverse transform) (1 1 10), a second inverse transform determining unit (or an element for determining the second inverse transform) ( 1 120), the second inverse transform unit (or to perform the second inverse transform 2019/190284 1 »（： 1 ^ 1 {2019/003743

32 elements) 1 130, and a first inverse transform unit (or an element that performs a first inverse transform) 1 140.

The second inverse transform application determining unit 1 1 10 may determine whether to apply the second inverse transform. For example, the secondary inverse transform may be a Non-Separable Secondary Transform (hereinafter referred to as NSST) or Reduced Secondary Transform (hereinafter referred to as RST). For example, the second inverse transform application determiner 1 1 10 may determine whether to apply the second inverse transform based on the second transform flag received from the encoder. As another example, the second inverse transform determining unit 1 1 10 may determine whether to apply the second inverse transform based on the transform coefficient of the residual block.

The second inverse transform determiner 1 120 may determine the second inverse transform. In this case, the second inverse transform determiner 1 120 may determine the second inverse transform applied to the current block based on the NSST (or RST) transform set specified according to the intra prediction mode.

In addition, as an embodiment, the secondary transform determination method may be determined based on the primary transform determination method. Various combinations of primary and secondary transforms may be determined according to the intra prediction mode.

Also, as an example, the second inverse transform determiner 1 120 may determine an area to which the second inverse transform is applied based on the size of the current block.

The second inverse transform unit 1 130 may perform a second inverse transform on the residual quantized residual block by using the determined second inverse transform.

The first inverse transform unit 1140 may perform a first inverse transform on the residual block that is second inverse transformed. The primary transform may be referred to as a primary transform or a core transform. By way of example, the first inverse transform unit 1140 may use the above-described MTS to perform the first order. 2019/190284 1 »（： 1 ^ 1 {2019/003743

33 Conversion can be performed. Also, as an example, the first inverse transform unit 1140 may determine whether MTS is applied to the current block.

For example, when the MTS is applied to the current block (that is,

The first inverse transform unit 1 140 based on the intra prediction mode of the current block.

Candidates can be configured. For example, the 1 ^ 8 candidate may consist of a combination of 0814 and / or DCT4 ^ \, or may include a combination of 0817 and / or DCTSS]. Alternatively, the MTS candidate may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and 27 to be described later.

And, the first inverse transform unit 1140 is configured

Particular candidates

The 11 _ \ indicating the MTS may be used to determine the primary transform applied to the current block.

Each of the above embodiments may be used, but the present invention is not limited thereto and may be used in combination with the above embodiments and other embodiments of the present specification. 12 is a block diagram for performing inverse transformation based on transformation related parameters in an embodiment to which the present invention is applied.

The decoder 200 to which the present invention is applied includes an element 1210 for acquiring sequence parameters, a multi-conversion selection flag (element 1220 for acquiring a power factor), and a multi-conversion selection index (^ ᄄ output 0 ^ _} 0 rule. Element 1230 to obtain and element 1240 to derive the transform kernel.

The element 1210 obtaining the sequence parameter may acquire sps_mts_intra_enabled_flag or sps_mts_intcr_enabled_flag. here

whether tu_mts_flag exists in the residual coding syntax of the intra coding unit, and sps_mts_inter_enabled_flag indicates whether tu_mts_flag exists in the residual coding syntax of the inter coding unit. As a specific example, the description of FIG. 12 may be applied.

An element 1220 that obtains a MTS flag may obtain tu_mts_flag based on sps_mts_intra_enabled_flag or sps_mts_inter_enabled_flag. For example, when sps_mts_intra_enabled_flag = 1 or sps_mts_inter_enabled_flag = 1, the element 1220 obtaining the MTS flag may acquire tu_mts_flag. Here, tu_mts_flag indicates whether multi-transform selection is applied to the residual sample of the luma transform block. As a specific example, the description of FIG. 12 may be applied.

An element 1230 that obtains a MTS index may obtain mts_idx based on tu_mts_flag. For example, when tu_mts_flag = 1, the element 1230 for obtaining the MTS index may obtain mts_idx. Here, mts_idx indicates which transform kernel is applied to luma residual samples along the horizontal and / or vertical direction of the current transform block. For example, at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and / or 27 described later may be applied.

An element 1240 for deriving a transform kernel may derive a transform kernel corresponding to mtsjdx.

The decoder 200 may perform inverse transformation based on the transform kernel. 2019/190284 1 »（： 1/10 公 019/003743

35 The above embodiments may be used respectively, but the present invention is not limited thereto and may be used in combination with the above embodiments and other embodiments of the present specification. 13 is an embodiment to which the present invention is applied and shows a flowchart of performing inverse transformation based on transformation related parameters.

The decoder to which the present invention is applied may acquire sps_mts_intra_enabled_flag or sps_mts_inter_enabled_flag (S 1310). Here, sps_mts_intra_enabled_flag indicates whether tu_mts_flag exists in the residual coding syntax of the intra coding unit. For example, if sps__mts__intra_enabled_flag = 0, tu_mts__flag is not present in the residual coding syntax of the intra coding unit, and if sps_mts_intra_enabled_flag = 0, tu_mts_flag is present in the residual coding syntax of the intra coding unit. In addition, sps_mts_i nter_enab 1 ed_fl ag indicates whether tu__mts_flag exists in the residual coding syntax of the inter coding unit. For example, if sps_mts_inter_enabled_flag = 0, tu_mts_flag is not present in the residual coding syntax of the inter coding unit, and if sps_mts_inter_enabled_flag = 0, tu_mts_flag is present in the residual coding syntax of the inter coding unit.

The decoder may acquire tu_mts_flag based on sps_mts_intra_enabled_flag or sps_mts_inter_enabled_flag (S1320). For example, when sps__mts_intra_enabled_flag = 1 or sps_mts_inter_enabled_flag = 1, the decoder can acquire tu__mts_flag. Here, tu_mts_flag indicates whether multiple transform selection (hereinafter, referred to as 'MTS') is applied to the residual sample of the luma transform block. For example, if tu_mts_flag = 0, the MTS is luma \ ¥ 0 2019/190284 1 （1710 {2019/003743

36 Not applied to residual samples in transform block

If 1, MTS is applied to the residual sample of the luma transform block.

As another example, at least one of the embodiments of the present document may be applied to the tix_mtsJlag.

The decoder

Based on 11 伯 _1 (you can get urine 1330). For example, when _11 伯 _ 3§ = 1, the decoder

Can be obtained. Here, 0 indicates which transform kernel is applied to luma residual samples along the horizontal and / or vertical direction of the current transform block.

For example, for b, at least one of the document embodiments may be applied. Specifically, at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and / or 27 described later may be applied.

remind

A corresponding translation kernel can be derived 1340. For example, the 11__1 (conversion kernel corresponding to urine) may be defined by being divided into a horizontal transform and a vertical transform.

As another example, different transform kernels may be applied to the horizontal transform and the vertical transform. However, the present invention is not limited thereto, and the same transform kernel may be applied to the horizontal transform and the vertical transform.

The decoder may perform inverse transformation based on the transform kernel.

Each of the above embodiments may be used, but the present invention is not limited thereto and may be used in combination with the above embodiments and other embodiments of the present specification. 2019/190284 1 »（： 1 ^ 1 {2019/003743

37 is an embodiment to which the present invention is applied and shows an encoding flowchart of performing Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) to a forward DCT2 or an inverse DCT2. Indicates.

The encoder may determine (or select) a horizontal transform and / or a vertical transform based on at least one of a prediction mode, a block shape, and / or a block size of the current block (S1410). In this case, the candidate for horizontal transformation and / or vertical transformation may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and / or 27 to be described later.

The encoder may determine an optimal horizontal transformation and / or an optimal vertical transformation through RD optimization (Rate Distortion optirr ^ zation). The optimal horizontal transform and / or the optimal vertical transform may correspond to one of a plurality of transform combinations, and the plurality of transform combinations may be defined by a transform index.

The encoder may signal a transform index corresponding to the optimal horizontal transform and / or the optimal vertical transform (S1420). Here, other embodiments described herein may be applied to the conversion index. For example, it may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and 27 to be described later.

As another example, a horizontal transform index for the optimal horizontal transform and a vertical transform index for the optimal vertical transform may be signaled independently. The encoder may perform a forward transformation in the horizontal direction with respect to the current block by using the optimal horizontal transformation (S1430). Here, the current block may mean a transform block. 2019/190284 1 »（： 1 ^ 1 {2019/003743

In operation S1440, the encoder may perform a forward transform in a vertical direction with respect to the current block by using the optimal vertical transform. In the present embodiment, the vertical transformation is performed after the horizontal transformation, but the present invention is not limited thereto. That is, the vertical transformation may be performed first, and then the horizontal transformation may be performed. In an embodiment, the forward direction in the horizontal forward direction conversion of the step S1430

DST4 is applied, and the forward direction in the vertical forward transform of step S1440

DCT4 can be applied. Or vice versa.

In one embodiment, the combination of the horizontal transform and the vertical transform may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and 27 to be described later.

Meanwhile, the encoder may generate a transform coefficient block by performing quantization on the current block (S 1450).

The encoder may perform entropy encoding on the transform coefficient block to generate a bitstream. FIG. 15 shows a decoding flowchart of performing _¾ DST4 (Discrete Sine Transform-4) and DCT4 (Discrete Cosine Transfomi-4) as a forward DCT2 or a reverse DCT2 as an embodiment to which the present invention is applied. .

The decoder may obtain a transform index from the bitstream (S1510). Here, other embodiments described herein may be applied to the conversion index. For example, it may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and 27 to be described later. 2019/190284 1 »（： 1 ^ 1 {2019/003743

The decoder may induce a horizontal transform and a vertical transform corresponding to the transform index 1520. In this case, the candidate for the horizontal transformation and / or the vertical transformation may include at least one of the embodiments of FIG. 6 and the embodiments of FIGS. 26 and / or 27 to be described later.

However, steps 81510 and 81520 are an embodiment, and the present invention is not limited thereto. For example, the decoder may induce a horizontal transform and a vertical transform based on at least one of a prediction mode, a block shape, and / or a block size of the current block. As another example, the transform index may include a horizontal transform index corresponding to a horizontal transform and a vertical transform index corresponding to a vertical transform.

Meanwhile, the decoder may entropy decode the bitstream to obtain a transform coefficient block and perform inverse quantization on the transform coefficient block (81530). The decoder may perform inverse transformation in the vertical direction on the inverse quantized transform coefficient block by using the vertical transformation.

The decoder may perform inverse transformation in the horizontal direction using the horizontal transformation (81550).

In the present embodiment, the horizontal transformation is applied after the vertical transformation is applied, but the present invention is not limited thereto. That is, the horizontal transformation may be applied first, and then the vertical transformation may be applied.

In an embodiment, the reverse direction 14 is applied in the vertical reverse direction conversion of the step 81540, and the reverse å (the 4) may be applied in the horizontal direction reverse conversion of the 81440 step, or vice versa.

In one embodiment, the combination of the horizontal transform and the vertical transform is shown in FIG. 2019/190284 1 »（： 1 ^ 1 {2019/003743

40 embodiments, which may include at least one of the embodiments of FIGS. 26 and 27 to be described later.

The decoder generates a residual block through step S1550, and the reconstructed block is generated by adding the residual block and the prediction block. FIG. 16 shows diagonal elements for a pair of transform block size (N) and right shift amount (SO) when DST4 and DCT4 are performed in forward DCT2 as an embodiment to which the present invention is applied.

The present invention proposes a method of reducing memory usage and computational complexity for Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) among transform kernels for video compression.

In one embodiment, the present invention provides a method of performing Discrete Sine Transform-4 (DST4) and Discrete Cosine Transform-4 (DCT4) to forward DCT2.

In one embodiment, a method of performing DST4 and DCT4 to inverse DCT2 is provided.

In one embodiment, a method of applying DST4 and DCT4 to a transform configuration group to which MTS (Multiple Transform Selection) is applied is provided. Example 1 design of DST4 and DCT4 with DCT2

The equation for inducing the matrices of DST4 and DCT4 is as follows.

[Equation 4] 2019/190284 1 »(： 1/10 公 019/003743

41

Here, n (0, ... N-1) represents a row index, and k (0, ... N-1) represents a column index. In this case, Equations 4 and 5 generate inverse transform matrices of DST4 and DCT4, respectively. And these transposes represent a forward transformation matrix.

If the DST4 (DCT4) inverse transform matrix is represented by (成¹ ((C ^)), the relations of the following equations 6 and 7 can be confirmed.

[Equation 6]

According to Equations 6 and 7, the present invention changes the input order or the output order through a pre-processing stage or a post-processing stage and signs. By changing the DCT4 (DST4) inverse transform matrix \ 02019/190284 1 »(： 1/10 公 019/003743

42

From (S ^) (((才)), we can derive the inverse transform matrix (<) ((()), respectively.

Therefore, the present invention can easily derive one from another without additional calculations when performing 0 4 or 00141 −.

In one embodiment of the present invention, £> (# 4 is £) (# 2) can be expressed as follows.

[Equation 8]

_{^{! (C n v J = (}} C; I) = A N (C:!! JM

Where is the post-processing matrix, and 山 is the pre-processing matrix.

For example,

In the present invention, it can be seen from Equation 8 that 1X ^ 4 can be designed based on the post-processing matrix 71, the preprocessing matrix, and 1X72. here, 2019/190284 1 »(： 1/10 公 019/003743

43 For post and preprocessing matrices only a small amount of multiplication is added. And, DCT2 can reduce the number of coefficients to be stored and is well known as a transformation for fast implementation based on symmetry between coefficients in DCT2 matrix.

Therefore, by adding a few multiplying factors, it is possible to realize high speed execution of £ ⁾ (刀 4 ^{) with} low complexity, even in the case of 0 and 14.

The inverse of the post-processing matrix ᆻ and the preprocessing matrix ᆻ can be expressed by Equation 9 below.

[Equation 9]

By using A − ′ of Equation 9, the present invention provides 1 ⁾ 4 and 0012. 2019/190284 1 »(： 1/10 公 019/003743

Another relationship, such as the following equation 10 between 44, can be derived.

[Equation 10]

[cfh [C = M- cfjA _N- '

Here, A- ^] , M ^~] are composed of simpler multiply than (), it is possible to fast implementation of DCT4 with low complexity. And A- 'results in fewer additions or subtractions than and v, but the coefficients in MJ- have a wider range than M _N. Accordingly, the present invention can design a conversion type based on Equations 9 and 10 in consideration of a tradeoff between complexity and performance.

From Equations 7, 8 and 10, the present invention can execute low complexity DST4 by reusing a fast implementation of DCT2. This is shown in the following equations (11) and (12).

[Equation 1 1]

（To） = （섟 ，） = （幻）} ·）

2019/190284 1 »(： 1/10 公 019/003743

45

Example 2 Implementation of DST4 and DCT4 with Forward DCT2

When Equation 11 is used to execute DST4, the input vector of length N must first be scaled by M _N J _n ). Similarly, when Equation 8 is used for the execution of DCT4, the input vector of length N must first be scaled by (M _n ).

Diagonal elements in M _N are floating point numbers, which must be scaled appropriately for use in fixed-point or integer multiplications. Integerized (M _N J _n ) and

(M _Nj J and

If you display

And ik /; can be calculated according to the following equations (13).

[Equation 13]

2019/190284 1 »(： 1/10 公 019/003743

46

16 shows M _N 'examples based on # and 및. Here, diag (·) means converting the argument matrix into an associative vector constituting the diagonal elements in the argument matrix. Acid _((%, and) of the same (ss幻) can easily be derived from the FIG. 16 by changing the order of the elements in each vector. For example, [251,213,142,50] may be changed to [50,142,213,251].

The present invention may set 5 / differently for each. For example, it may be set to 7 for a 4x4 transform and 5; to 8 for an 8x8 transform. Of Equation 13

Denotes the amount of left shift to scale by 2 " ¹ , and the" skin line "operator performs the appropriate 1 ^> 01111 (111. 八 and () are diagonal matrices and the 1 * elements of the input vector X (<11 ( （3 公） is multiplied by [好，, Multiplication result and diagonal of the input vector X

The matrices can be expressed as Equation 14 below.

【Equation ₁₄ 】

Fig. 14 shows the multiplication result. X must be scaled down later. Downscaling of ^ is performed before applying 1X72, or 2019/190284 1 »(： 1/10 公 019/003743

47

0 ⁽ performed after applying 2, or after multiplying 山 ((公_; 0) by 1 × 74 肝 4). If the scaled down of the figure is performed before applying DCT2, the scaled down ( the down-scaled one), x may be determined based on Equation 15 below.

[Equation 15]

In Equation 15, ¾ may be the same value as described above. However, the present invention is not limited thereto, and ¾ may have a different value from the above.

In Equation 15, any type of scaling and rounding may be used, and (1) and (2) of Equation 15 may be used in one embodiment. That is, as shown in Equation 15, (1), (2) or other functions may be applied to find a station. 17 and 18 are embodiments to which the present invention is applied, and FIG. 17 shows sets of DCT2 kernel coefficients that can be applied to DST4 or DCT4, and FIG. 18 shows a forward DCT2 matrix generated from a set of DCT2 kernel coefficients.

One embodiment of the present invention may use the same DCT2 kernel coefficients as HEVC. It is necessary to maintain 31 different coefficients of DCT2, which is facilitated by symmetries among all DCT2 kernel coefficients of all sizes up to 32x32, used by symmetry between all DCT2 kernel coefficients of up to 32x32.

If you reuse an existing implementation of DCT2, DST4 or DCT4 2019/190284 1 »（： 1 ^ 1 {2019/003743

There is no need to store additional coefficients of DCT2 used at 48.

If a specific DCT2 kernel is used rather than an existing DCT2, the present invention may add only one set of DCT2 kernel coefficients, which are 31 coefficients using homogeneous symmetry. That is, if up to 2 ⁿ x 2 ⁿ DCT2 is supported, the present invention only needs (2 ⁿ -l) different coefficients.

This additional set may have higher or lower accuracy than before. If the dynamic range of x does not exceed the range supported by the existing DCT2 design, the present invention will reuse the same routine as DCT2 without extending the bit length of internal variables. And reuse the legacy design of DCT2.

Although it requires more arithmetic accuracy of DST4 / DCT4 than DCT2, an updated routine that can accumulate higher accuracy is also sufficient to perform existing DCT2. For example, more accurate sets of DCT2 coefficients are listed in FIG. 17 above according to scaling factors.

Each coefficient of FIG. 17 may be further adjusted to improve orthogonality between basis vectors, the norm of each basis vector is close to 1, and the Frobenius norm error can be reduced from a floating-point accurate DCT2 kernel. . If the coefficient set is (a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t, u 八 w, x, y, z, A, B, C, D, E), the forward DCT2 generated from the coefficient set can be set as shown in FIG.

In FIG. 18, each DCT2 coefficient set (each row in FIG. 18) is represented by (a, b _? C, d, e, f, g, h, i, j, k, l, m, n, o, p , q, r, s, t, u, v, w, x, y, z, A, B, C, D, E). This is 31 2019/190284 1 »（： 1 ^ 1 {2019/003743

Only 31 possibly different coefficients reflect the need for all DCT2 transforms of size not greater than 32x32.

The output of the DCT2 transform must be post-processed via the matrix A _N (or DNA _N ) of DCT4 (or DST4). Prior to providing an input vector to the matrix A _N (or to the ratio of the DCT4 (or DST4), the DCT2 output vector as the input vector may be rounded to a value for accuracy adjustment to store as variables of defined bit lengths. Assuming that the DCT2 output vector before scaling and rounding is y, a rounded one y can be determined from the following equation (16).

As with 15, other forms of scaling and rounding are applicable to Equation 16 above.

[Equation 16]

In Equation 16, when S3 is 0, no scaling and rounding are applied to S ,. That is, X = y _f .

Let X be the final output vector after multiplying A _N or (£ ⁾ ^ by. Most of the multiplication can be replaced by simple addition or telling, except for the first 1 / V2 multiplication. Here, since the 1 / V2 factor is a constant number, it can be approximated by hardwired multiplication, that is, shift, as shown by Equation 17 below. As in equation (15), other forms of scaling and rounding are applicable to equation (17).

[Equation 17]

x _0- _(y0 -F _{+ 0 «} (¾-1)))» ¾ 2019/190284 1 »（： 1/10 公 019/003743

In Equation 17 above, and are satisfied that F »S4 is very close to 1 / V2. One way to obtain (F, S4) pairs is by using F = roundel V2) «S _A }.

Although the present invention can increase the public for a more accurate approximation to 1 / V2, an increase in S4 requires intermediate variables of longer length, which increases execution complexity. You can. Table 1 below shows possible pairs of (F, S4) approximating 1 / V2.

Table 1

In Equation 17, it is assumed that the present invention applies a right shift equal to the left shift of in order not to change the overall scaling, but need not be. If, when applying the right shifted by ¾ (<S ₄₎ instead of S _4, the present invention is

All j) must be scaled up. Considering all the shifts of the previous equations and the expected result scaling after DCT4 (or DST4) calculation (5V, where positive value means right shift), the present invention provides all scaling bit shift values ( Equation 18 with all the scaling bit shift values) can be set. 2019/190284 1 »(： 1 ^ 1 {2019/003743

51

[Equation 18]

S _r = (S ^-S ₂ ) + S _c -S ₃ + (S ₄ -S ₅ )-S ₀

In Equation 18, Sc represents a left shift amount due to DCT2 integer multiplication, which may be a non-integer value as shown in FIG. 17. So represents the right shift amount that yields the final output of DCT4 (or DST4). Some portions of Equation 18 may be zero. For example, (5 /-&), ¾ or (Ss-S ₄ ) can be zero. 19 and 20 illustrate embodiments to which the present invention is applied. FIG. 19 illustrates code execution of an output stage for DST4, and FIG. 20 illustrates code execution of an output stage for DCT4.

Speaking of the i th element of the final output vector, an embodiment of the present invention can provide an example of final step code execution for DST4 corresponding to multiplication of the public as shown in FIG.

In addition, another embodiment of the present invention may provide an example of final step code execution for DCT4 corresponding to multiplication of as shown in FIG.

The cutoff in FIG. 19 represents the effective number of coefficients in the vector X. For example, the cutoff may be N.

In FIG. 19, steps S 1910 and S1920 may be merged into one calculation process as shown in Equation 19 below.

[Equation 19]

X。 = [7 江 »3 ( _£;次« / »« " _{? ¾} , _£ / ½3)«> «·, (% F + (1« ( _> 5 ₅ + 5-1))) »(5 + ¾)) 2019/190284 1 »（： 1 ^ 1 {2019/003743

52 As in Equation 15 above, other forms of scaling and rounding are applicable to FIG. 19 and Equation 19 above.

In FIG. 20, steps S2010 and S2020 may be merged into one calculation process as shown in Equation 20 below.

[Equation 20]

X ₀ -Clip3 (, clipMinimum, clipMaximwn, (y ₀ -F + (1 «(S ₅ + S _0-1 ))» (S ₅ + S _Q )) As shown in Equation 15 above, scaling and rounding Shapes are also applicable to FIG. 20 and Equation 20 above.

In FIG. 19 and FIG. 20, Clip3 represents an operation of clipping the argument value to both ends (clipMinimum, clipMaximum).

A _N (or (D _N A _n )) ^ \ Each row has a common pattern with its previous row, and the present invention is directed to appropriate sign sign reversal. You can reuse a result of the previous row. Such a pattern may be utilized through the variables z and prev in FIGS. 19 and 20. Here, the variable z, prev reduces the multiplication calculation of A _N (or (公 ᆻ)).

With the variables z and prev, the present invention requires only one multiplication or one addition / subtraction for each output. For example, multiplication may be necessary only at the first output element. FIG. 21 illustrates an embodiment to which the present invention is applied and shows configuration of parameter sets and multiplication coefficients for DST4 and DCT4 when DST4 and DCT4 are performed in forward DCT2. 2019/190284 1 »（： 1 ^ 1 {2019/003743

53 shows the configuration of parameter sets and multiplication coefficients for DST4 and DCT4 of DST 4 and DCT4. Each transform of a different size can be configured separately. In other words, each transform of a different size may have its own parameter set and multiplication coefficients.

For example, when the configuration of the parameter set of DST4 is (Si, S2, Ss, S4, S5, So), the multiplication factor value is (8, 8, 0, 8, 8, identical to HEVC for all block sizes). Can be). And, when the configuration of the parameter set of DCT4 is (Si, S ₂ , S ₃ , S4, Ss, So), the multiplication factor value is (8, 8, 0, 8, 8, identical to HEVC for all block sizes). Can be).

Further, when the configuration of the parameter set is M _N , each block size may have respective multiplication coefficient values described in FIG. 21.

According to the present invention, according to Equation 18, execution of the reverse DST4 [DCT4] is the same as the forward DST4 [DCT4]. 22 and 23 illustrate embodiments to which the present invention is applied. FIG. 22 illustrates code execution of a preprocessing step for DCT4, and FIG. 23 illustrates code execution of a preprocessing step for DST4.

Example 3: Alternative implementation of DST4 and DCT4 with inverse DCT2 The present invention provides a method of executing DCT4 and DST4 through Equations 10 and 12, respectively.

A, (A _N -'J _n ), M- \ and (D _N MJ-) may be used instead of A _n , (D _N A _n ), M _n , and {M _N J _n , each of which Requires less computation compared to DCT2. Reverse DCT2 is applied instead of forward DCT2 in Equations 10 and 12 above. 2019/190284 1 »（： 1/10 公 019/003743

54. In contrast to the equation (8) and 11, A- ', or (A ^~' J _n) is applied at the input vector is applied on the x, M- 'or (D MJ- _N) is the output vector of DCT2.

As in Equations 9 and 12, only one element is multiplied by V2 in AJ- and (^ ¹ ^). Here, A 'and (A ^~] J _n ) may be approximated by integer multiplication by the right shift.

An example of code execution in the preprocessing step of DCT4 in Equation 10 is shown in FIG. 22, which corresponds to the multiplication of A ^˜ ′. In addition, an example of code execution in the preprocessing step of DST4 in Equation 12 is shown in FIG. 23, which corresponds to (multiplication of three times).

As in Equation 15, other forms of scaling and rounding are applicable to the following Tables 8 and 9.

In FIGS. 22 and 23, N indicates the length of the transform basis vector as well as the length of the input vector x. F and 5 ¹ / silver, relation

The multiplication factor and the right shift amount to approximate V are shown.

22 and 23, since the input vector should be scaled up by 2 ^S1 _ ^S2 , S2

Used for rounding instead of Si. If Si is equal to Si, there is no need to scale the input vector. Table 2 below shows an example of the (F, S0 pair) to approximate the anti-multiplication.

Table 2

2019/190284 1 »(： 1/10 公 019/003743

55

As in Equation 16, the present invention can scale down the reverse 1X ^ 2 output to take advantage of shorter bit length variables. If the backward 1X ^ 2 output vector is ^ and the W element is written, the scaled output vector can be obtained according to Equation 21 below. As in equation (15), other forms of scaling and rounding are applicable to equation (21).

[Equation 21]

女 = (刀 + (1 «(¾-1))) ñ ñ 公₃ ， _I = 0,1, ..., ^-1

In Equations 10 and 12, the post-processing steps correspond to and (· å>), respectively. Here, the associated diagonal coefficients can be scaled up for fixed point or integer multiplication. This scale up may be performed with appropriate left shifts as shown in Equation 22 below.

[Equation 22]

24 is an embodiment to which the present invention is applied, and 0814 and 1X74 are reverse 0012. 2019/190284 1 »(： 1 ^ 1 {2019/003743

Diagonal elements for a pair of transform block sizer) and right shift amount 4) when executed at 56. Examples of diagonal elements of can be seen in various combinations of Figures 24 and VII above.

As in Embodiment 2, S ₄ may be set differently for each transform size. In FIG. 24, when (N, S4) is (32, 9), large numbers such as '10431,' can be decomposed as shown in Equation 23 below, which is suitable for multiplication of a shorter bit length operator part. This can be applied when a large number of multiplications appear.

[Equation 23]

10431-x = (8096 + 2048 + 287) x = (JC «13) + (x« 1 1) + (287x)

Corresponding examples can be derived from FIG. 24 above. For example, when (N, S4) is (4, 9), the vector is [261, -308, 461, -1312].

Non-zero elements are only available on diagonal lines in M and {D _N M ^), and the associated matrix multiplication is element-wise multiplication, as in Equation 24 It can be performed by).

[Equation 24]

If the final output vector is X, 1 calculated from Equation 24 is previously 2019/190284 1 »（： 1 ^ 1 {2019/003743

57 It must be scaled appropriately to satisfy the given example scaling. For example, if the left shift amount for obtaining the final output vector X is So and the predicted scaling is ST, the overall relationship between the shift lengths along with 切 and 는 is expressed as Can be set.

[Equation 25]

X, = (i-, + (1 «(¾-1)))» ¾, z = 0,1, ..., iV-1

Here, ST may have a non-negative value as well as a negative value. Sc may have the same value as in Equation 18. As with Equation 15, other forms of scaling and rounding are applicable to Equation 25. FIG. 25 illustrates a configuration of parameter sets and multiplication coefficients for DST4 and DCT4 when DST4 and DCT4 are performed in reverse DCT2 as an embodiment to which the present invention is applied.

25 illustrates the configuration of parameter sets and multiplication coefficients in alternative implementations for DST4 and DCT4. Each transform of a different size can be configured separately. In other words, each transform of a different size may have its own parameter set and multiplication coefficients.

For example, when the component of the parameter set of DST4 (Si, S ₂ , S3, S4, Ss, So), the multiplication factor value for all block sizes is (8, 8, 0, 8, 8, identical to HEVC). 2019/190284 1 »（： 1 ^ 1 {2019/003743

58 And, when the configuration of the parameter set of DCT4 is (Si, S ₂ , S ₃ , S ₄ , S ₅ , So), for all block sizes, the multiplication coefficients are (8, 8, 0, 8, 8, identical to HEVC). In addition, when the configuration of the parameter set is MJ ^~ , each block size may have respective multiplication coefficient values described in FIG. 25.

According to the present invention, according to Equation 18, the execution of the reverse DST4 [DCT4] is the same as the forward DST4 [DCT4]. 26 and 27 illustrate embodiments to which the present invention is applied, FIG. 26 shows MTS mapping for intra prediction residual, and FIG. 27 shows inter prediction residual.

Represents MTS mapping. Embodiment 4 Possible Multiple Transform Selection (MTS) mapping with DST4 and DCT4 In one embodiment of the invention, DCT4 and DST4 may be used to generate MTS mapping. For example, DST7 and DCT8 may be replaced by DCT4 and DST4.

In another embodiment, only DCT4 and DST4 may be used to generate the MTS. For example, the following Tables 13 and 14 show MTS examples for intra predicted residual and inter predicted residual, respectively.

In another embodiment of the present invention, mapping is also possible by other combinations of DST4, DCT4, DCT2, and the like.

In another embodiment, an MTS configuration that replaces DCT4 with DCT2 is possible. 2019/190284 1 »（： 1 ^ 1 {2019/003743

In another embodiment, the mapping for the inter predicted residual consisting of 1 × 78/0817 may be retained and replaced only for the intra predicted residual. In another embodiment, a combination of the above embodiments is possible. 28 is a diagram illustrating a structure of a content streaming system according to an embodiment to which the present invention is applied.

Referring to FIG. 28, a content streaming system to which the present invention is applied may largely include an encoding server, a streaming server, a web server, a media storage, a user device, and a multimedia input device.

The encoding server compresses content input from multimedia input devices such as a smartphone, a camera, a camcorder, etc. into digital data to generate a bitstream and transmit the bitstream to the streaming server. As another example, when multimedia input devices such as smart phones, cameras, camcorders, etc. directly generate a bitstream, the encoding server may be omitted.

The bitstream may be generated by an encoding method or a bitstream generation method to which the present invention is applied, and the streaming server may temporarily store the bitstream in the process of transmitting or receiving the bitstream. The streaming server transmits the multimedia data to the user device based on the user's request through the web server, and the web server serves as a medium for informing the user of what service. When a user requests a desired service from the web server, the web server delivers it to a streaming server, and the streaming server transmits multimedia data to the user. At this time, 2019/190284 1 »（： 1 ^ 1 {2019/003743

60 The content streaming system may include a separate control server. In this case, the control server controls a command / response between devices in the content streaming system.

The streaming server may receive content from a media store and / or an encoding server. For example, when the content is received from the encoding server, the content may be received in real time. In this case, in order to provide a smooth streaming service, the streaming server may store the bitstream for a predetermined time.

Examples of the user device include a mobile phone, a smart phone, a laptop computer, a digital broadcasting terminal, a personal digital assistant (PDA), a portable multimedia player (PMP), a navigation device, a slate PC, Tablet PCs, ultrabooks, wearable devices such as watchwatches, glass glasses, head mounted displays, digital TVs, desktops Computer, digital signage, and the like.

Each server in the content streaming system may be operated as a distributed server, in which case data received from each server may be distributed. As described above, the embodiments described herein may be implemented and performed on a processor, microprocessor, controller, or chip. For example, the functional units shown in each drawing may be implemented and performed on a computer, a processor, a microprocessor, a controller, or a chip.

In addition, the decoder and encoder to which the present invention is applied are multimedia broadcast transmission and reception 2019/190284 1 »（： 1 ^ 1 {2019/003743

61 devices, mobile communication terminals, home cinema video devices, digital cinema video devices, surveillance cameras, video chat devices, real time communication devices such as video communication, mobile streaming devices, storage media, camcorders, video on demand (VoD) services , OTT video (Over the top video) device, Internet streaming service providing device, three-dimensional (3D) video device, video telephony video device, and medical video device, etc., can be used to process the video signal or data signal have. For example, the OTT video (Over the top video) device may include a game console, a Blu-ray player, an Internet-connected TV, a home theater system, a smartphone, a tablet PC, a digital video recorder (DVR), and the like.

In addition, the processing method to which the present invention is applied can be produced in the form of a program executed by a computer, and stored in a computer-readable recording medium. Multimedia data having a data structure according to the present invention can also be stored in a computer-readable recording medium. The computer readable recording medium includes all kinds of storage devices and distributed storage devices in which computer readable data is stored. The computer-readable recording medium may be, for example, a Blu-ray disc (BD), a universal serial bus (USB), a ROM, a PROM, an EPROM, an EEPROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, and an optical disc. It may include a data storage device. The computer-readable recording medium also includes media embodied in the form of a carrier wave (for example, transmission over the Internet). In addition, the bitstream generated by the encoding method may be stored in a computer-readable recording medium or transmitted through a wired or wireless communication network.

In addition, an embodiment of the present invention is a computer program by the program code 2019/190284 1 »（： 1 ^ 1 {2019/003743

62 may be implemented in a product, and the program code may be executed on a computer by an embodiment of the present invention. The program code may be stored on a carrier readable by a computer.

Industrial Applicability

Above, a preferred embodiment of the present invention described above is, ^as' described for illustrative purposes, those skilled in the art within the technical less spirit and the technical scope of the invention as set forth in the claims hereinafter appended, improved various other embodiments Changes, substitutions or additions will be possible.

Claims

2019/190284 1 »（： 1 ^ 1 {2019/003743 63 Scope of claim]

[Claim 1]

A method of reconstructing a video signal based on low complexity conversion execution, the method comprising: obtaining a transform index of a current block from the video signal, wherein the transform index is one of a plurality of transform combinations consisting of a combination of 0814 and / or 0014 Corresponds to either one;

Deriving a transform combination corresponding to the transform index, wherein the transform combination comprises a horizontal transform and a vertical transform, wherein the horizontal transform and the vertical transform correspond to either 0814 or 0014;

Performing an inverse transform in a vertical direction with respect to the current block using the 14;

Performing an inverse transform in the horizontal direction with respect to the current block using the 1 ^{) (} 14 ⁾ ; and

Reconstructing the video signal using the inverted current block

Method comprising a.

[Claim 2]

The method of claim 1,

0814 and / or the DCT4³ forward 1X72 or reverse

Characterized in that it is executed using.

[Claim 3]

The method of claim 2, 2019/190284 1 »（： 1/10 公 019/003743

64, characterized in that 14 and / or 1) (where 4 is the forward 1) 2 or the reverse post-processing matrix and preprocessing matrix wool.

Size)

[Claim 4]

The method of claim 1,

And applying the inverse transform of the column for each column when the vertical transform is 14, and applying the inverse transform of 1 × 714 for each row when the horizontal transform is 1 × 714.

[Claim 5]

The method of claim 1,

The conversion combination (horizontal transformation, vertical transformation) is 4, 14), (1X74, 14),

Method comprising a.

[Claim 6]

The method of claim 5,

And when the current block is an intra predicted residual, the transform combination corresponds to the transform indices 0, 1, 2, and 3.

[Claim 7] 2019/190284 1 »（： 1 ^ 1 {2019/003743

65 The method of claim 5,

And when the current block is an inter predicted residual, the transform combination corresponds to the transform index 3, 2, 1, 0.

[Claim 8]

An apparatus for reconstructing a video signal based on low complexity conversion execution, the apparatus comprising: a parser for obtaining a transform index of a current block from the video signal, wherein the transform index is a plurality of transform combinations composed of a combination of 0814 and / or Dm. Corresponds to any one of;

A transform unit for deriving a transform combination corresponding to the transform index, performing an inverse transform in the vertical direction with respect to the current block using 0 14, and performing an inverse transform in the horizontal direction with respect to the current block using 1X74 Wherein the transform combination comprises a horizontal transform and a vertical transform, wherein the horizontal transform and the vertical transform correspond to either 0814 or DCT4;

A reconstruction unit reconstructing the video signal using the inversely converted current block

Apparatus comprising a.

[Claim 9]

The method of claim 8,

0814 and / or £) (4 is forward 1X72 or reverse

Device, characterized in that it is executed using.

[Claim 10]

The method of claim 9, 2019/190284 1 »（： 1/10 公 019/003743

66 the 0814 and / or the ： 0 (where 4 is the forward £) a2 or the reverse

Post-processing matrix and pre-processing matrix are applied to 1X72

Size)

[Claim 11]

The method of claim 8,

And the inverse transform of 14 is applied for each column when the vertical transform is 14, and the inverse transform of 1X ^ 4 is applied for each row when the horizontal transform is 0 (4).

[Claim 12]

The method of claim 8,

The conversion combination (horizontal transformation, vertical transformation) includes 必肝 4, 0814, (1) 4, 14), (0814,0 4), and (0 4,1) 014. .

[Claim 13]

The method of claim 12,

And when the current block is an intra predicted residual, the transform combination corresponds to the transform index 0, 1,2,3.

[Claim 14] 2019/190284 1 »（： 1/10 公 019/003743

67 The method of claim 12,

And when the current block is an inter predicted residual, the transform combination corresponds to the transform index 3, 2, 1, or 0.